my $term = new Term::ReadLine 'yaz-cleanup';
my $OUT = $term->OUT || \*STDOUT;
+$| = 1;
+
# initialization and setup
my $conf = {};
initialize($conf);
-populate_trash() if ($conf->{trash});
+populate_trash() if ($conf->{trashfile});
# set up files, since everything appears to be in order
-open MARC, '<:utf8', shift || 'incoming.marc.xml';
-open my $NUMARC, '>:utf8', $conf->{output};
-open my $EXMARC, '>:utf8', $conf->{exception};
-open my $OLD2NEW, '>', 'old2new.incoming.map'
+open MARC, '<:utf8', (shift || 'incoming.marc.xml')
+ or die "Can't open input file $!\n";
+open my $NUMARC, '>:utf8', $conf->{output}
+ or die "Can't open output file $!\n";
+open my $OLD2NEW, '>', 'old2new.map'
if ($conf->{'renumber-from'} and $conf->{'original-subfield'});
-print $NUMARC '<?xml version="1.0" encoding="UTF-8"?>',"\n";
-print $NUMARC '<collection xmlns="http://www.loc.gov/MARC21/slim">',"\n";
+my $EXMARC = 'EX';
+
my @record = (); # current record storage
my %recmeta = (); # metadata about current record
}
do_automated_cleanups();
- stow_record_data();
$ptr = 0;
until ($ptr == $#record) {
}
write_record($NUMARC);
}
-print $NUMARC "</collection>\n";
+#print $NUMARC "</collection>\n";
print $OUT "\nDone. \n";
sub do_automated_cleanups {
$ptr = 0;
until ($ptr == $#record) {
+ # get datafield/tag data if we have it
+ stow_record_data();
+
# catch empty datafield elements
- if ($record[$ptr] =~ m/<datafield tag="..." ind1="." ind2=".">/) {
+ if ($record[$ptr] =~ m/<datafield tag="..."/) {
if ($record[$ptr + 1] =~ m|</datafield>|) {
my @a = @record[0 .. $ptr - 1];
my @b = @record[$ptr + 2 .. $#record];
sub stow_record_data {
# get tag data if we're looking at it
- if ($record[$ptr] =~ m/<datafield tag="(.{3})" ind1="(.)" ind2="(.)">/) {
- $recmeta{tag} = $1;
- $recmeta{ind1} = $2;
- $recmeta{ind2} = $3;
+
+ if ($record[$ptr] =~ m/<datafield tag="(?<TAG>.{3})"/) {
+ $recmeta{tag} = $+{TAG};
+ $record[$ptr] =~ m/ind1="(?<IND1>.)"/;
+ $recmeta{ind1} = $+{IND1} || '';
+ $record[$ptr] =~ m/ind2="(?<IND2>.)"/;
+ $recmeta{ind2} = $+{IND2} || '';
+
+ unless (defined $recmeta{tag}) {
+ message("Autokill record: no detectable tag");
+ dump_record("No detectable tag") ;
+ }
# and since we are looking at a tag, see if it's the original id
- if ($conf->{'original-subfield'} and $1 == $conf->{'original-tag'}) {
+ if ($conf->{'original-subfield'} and
+ $recmeta{tag} == $conf->{'original-tag'}) {
my $line = $record[$ptr]; my $lptr = $ptr;
my $osub = $conf->{'original-subfield'};
$recmeta{oid} = 'NONE';
until ($line =~ m|</record>|) {
$lptr++;
$line = $record[$lptr];
- $recmeta{oid} = $1
+ $recmeta{oid} = $+{TAG}
if ($line =~ /<subfield code="$osub">(.+?)</);
}
}
print_fullcontext();
# stow original problem line
- $conf->{origline} = $record[$ptr];
+ $recmeta{origline} = $record[$ptr];
while (1) {
my $line = $term->readline('marc-cleanup>');
my $term = $commands{$chunks[0]}->(@chunks[1..$#chunks]);
last if $term;
} else {
+ $recmeta{prevline} = $record[$ptr];
$record[$ptr] = "$line\n";
print_context();
}
my $l = '';
$l = <MARC> while (defined $l and $l !~ /<record>/);
return $l unless defined $l;
- @record = ($l);
+ @record = ();
%recmeta = ();
$conf->{ricount}++;
my ($FH) = @_;
my $trash = $conf->{trash};
+ if ($FH eq 'EX') {
+ $EXMARC = undef;
+ open $EXMARC, '>:utf8', $conf->{exception}
+ or die "Can't open exception file $!\n";
+ $FH = $EXMARC;
+ }
+
$conf->{rocount}++ if ($FH eq $NUMARC);
print $FH '<!-- ', $recmeta{explanation}, " -->\n"
if(defined $recmeta{explanation});
# add 903(?) with new record id
my $renumber = '';
if ($conf->{'renumber-from'}) {
- $recmeta{nid} = $conf->{'renumber=from'};
+ $recmeta{nid} = $conf->{'renumber-from'};
$renumber = join('', ' <datafield tag="', $conf->{'renumber-tag'},
'" ind1=" " ind2=" "> <subfield code="',
$conf->{'renumber-subfield'},
$conf->{'renumber-from'}++;
}
- # scrub newlines
- unless ($conf->{nocollapse})
+ # scrub newlines (unless told not to or writing exception record)
+ unless ($conf->{nocollapse} or $FH eq $EXMARC)
{ s/\n// for (@record) }
# write to old->new map file if needed
- print $OLD2NEW $recmeta{oid}, "\t", $recmeta{nid}, "\n"
- if ($conf->{'renumber-from'} and $conf->{'original-subfield'});
+ if ($conf->{'renumber-from'} and $conf->{'original-subfield'}) {
+ unless (defined $recmeta{oid}) {
+ my $msg = join(' ', "No old id num found");
+ dump_record($msg);
+ } else {
+ print $OLD2NEW $recmeta{oid}, "\t", $recmeta{nid}, "\n"
+ }
+ }
- # and finally, actually write the record
- print $FH @record;
+ # actually write the record
+ print $FH @record,"\n";
+
+ # if we were dumping to exception file, nuke the record and set ptr
+ # to terminate processing loop
+ @record = ('a');
+ $ptr = 0;
}
sub print_fullcontext {
print $OUT $conf->{editmsg},"\n";
print $OUT "\r Tag:",$recmeta{tag}, " Ind1:'",
$recmeta{ind1},"' Ind2:'", $recmeta{ind2}, "'";
+ print $OUT " @ ", $conf->{ricount}, "/", $conf->{rocount} + 1;
print_context();
return 0;
}
#-----------------------------------------------------------------------------------
sub substitute {
- my ($line_in, @chunks) = @_;
+ my (@chunks) = @_;
my $ofrom = shift @chunks;
- if ($ofrom =~ /^'/ or !@chunks) {
- until ($ofrom =~ /'$/)
+ if ($ofrom =~ /^'/) {
+ until ($ofrom =~ /'$/ or !@chunks)
{ $ofrom .= join(' ','',shift @chunks) }
$ofrom =~ s/^'//; $ofrom =~ s/'$//;
}
$from = join('', $from, $char);
}
- $conf->{prevline} = $record[$ptr];
+ $recmeta{prevline} = $record[$ptr];
$record[$ptr] =~ s/$from/$to/;
print_context();
return 0;
}
sub merge_lines {
- $conf->{prevline} = $record[$ptr];
+ $recmeta{prevline} = $record[$ptr];
# remove <subfield stuff; extract (probably wrong) subfield code
$record[$ptr] =~ s/^\s*<subfield code="(.*?)">//;
# and move to front of line
}
sub flip_line {
- unless ($conf->{prevline})
+ unless ($recmeta{prevline})
{ print $OUT "No previously edited line to flip\n"; return }
my $temp = $record[$ptr];
- $record[$ptr] = $conf->{prevline};
- $conf->{prevline} = $temp;
+ $record[$ptr] = $recmeta{prevline};
+ $recmeta{prevline} = $temp;
print_context();
return 0;
}
sub kill_line {
- $conf->{killline} = $record[$ptr];
+ $recmeta{killline} = $record[$ptr];
my @a = @record[0 .. $ptr - 1];
my @b = @record[$ptr + 1 .. $#record];
@record = (@a, @b);
}
sub yank_line {
- unless ($conf->{killline})
+ unless ($recmeta{killline})
{ print $OUT "No killed line to yank\n"; return }
my @a = @record[0 .. $ptr - 1];
my @b = @record[$ptr .. $#record];
return 0;
}
+sub insert_original {
+ $record[$ptr] = $recmeta{origline};
+ print_context();
+ return 0;
+}
+
sub display_lines {
- print $OUT "\nOrig. edit line :", $conf->{origline};
- print $OUT "Current flip line:", $conf->{prevline} if $conf->{prevline};
- print $OUT "Last killed line :", $conf->{killline} if $conf->{killline};
+ print $OUT "\nOrig. edit line :", $recmeta{origline};
+ print $OUT "Current flip line:", $recmeta{prevline} if $recmeta{prevline};
+ print $OUT "Last killed line :", $recmeta{killline} if $recmeta{killline};
print $OUT "\n";
return 0;
}
sub dump_record {
- my ($line_in, @explanation) = @_;
+ my (@explanation) = @_;
+ print $OUT @explanation;
$recmeta{explanation} = join(' ', 'Tag', $recmeta{tag}, @explanation);
write_record($EXMARC);
return 1;
sub populate_trash {
print $OUT ">>> TRASHTAGS FILE FOUND. LOADING TAGS TO BE STRIPPED FROM OUTPUT\n";
- open TRASH, '<', $conf->{trash}
+ open TRASH, '<', $conf->{trashfile}
or die "Can't open trash tags file!\n";
while (<TRASH>) {
my $lastwasrange = 0;
'autoscrub|a',
'exception|x=s',
'output|o=s',
+ 'prefix|p=s',
'nocollapse|n',
'renumber-from|rf=i',
'renumber-tag|rt=i',
- 'renumber-subfield|rs=i',
+ 'renumber-subfield|rs=s',
'original-tag|ot=i',
- 'original-subfield|os=i',
+ 'original-subfield|os=s',
'script',
'strip-nines',
- 'trash|t=s',
+ 'trashfile|t=s',
'trashhelp',
'help|h',
);
show_trashhelp() if ($c->{trashhelp});
# defaults
- $c->{output} = 'cleaned.incoming.marc.xml' unless defined $c->{output};
- $c->{exception} = 'cleaned.exceptions.marc.xml' unless defined $c->{exception};
+ if ($c->{prefix}) {
+ $c->{output} = join('.',$c->{prefix},'marc','xml');
+ $c->{exception} = join('.',$c->{prefix},'marc','ex');
+ }
$c->{'renumber-tag'} = 903 unless defined $c->{'renumber-tag'};
$c->{'renumber-subfield'} = 'a' unless defined $c->{'renumber-subfield'};
$c->{window} = 5;
Usage is: marc-cleanup [OPTIONS] <filelist>
Options
--output -o Cleaned MARCXML output filename
- (default: cleaned.incoming.marc.xml)
--exception -x Exception (dumped records) MARCXML filename
- (cleaned.exceptions.marc.xml)
- --trash File containing trash tag data (see --trashhelp)
+ or
+ --prefix=<PREFIX>> -p Shared prefix for output/exception files. Will
+ produce PREFIX.marc.xml and PREFIX.ex.xml
+
+ --trashfile -t File containing trash tag data (see --trashhelp)
- --renumber-from=NUM -rf Begin renumbering id sequence with this number
+ --renumber-from -rf Begin renumbering id sequence with this number
--renumber-tag -rt Tag to use in renumbering (default: 903)
--renumber-subfield -rs Subfield code to use in renumbering (default: a)
--original-tag -ot Original id tag; will be kept in output even if
it appears in the trash file
--original-subfield -os Original id subfield code. If this is specified
and renumbering is in effect, an old-to-new mapping
- file (old2new.incoming.map) will be generated.
+ file (old2new.map) will be generated.
--nocollapse -n Don't compress records to one line on output
--autoscrub -a Automatically remove non-numeric tags in data
projects / migration-tools.git / blobdiff